Airicist
2nd June 2016, 19:46
https://youtu.be/F_vb-oEs6T4
Matthew Travers : shape-based compliance in locomotion
Published on Jun 2, 2016
Matthew Travers
June 2, 2016
Abstract:
Natural systems move through unstructured, complex environments with an adaptability and robustness that motion-control scientists have yet to deeply understand, let alone reliably emulate. My work thus looks to biology, not to mimic or replay motions back on robotic devices, but to better understand the composition of successful motion-control strategies as they exist in the natural world. To this end, I will present recent work that uses modal decomposition techniques as well as tools from the field of geometric mechanics to better represent and ultimately interpret the motion of biological systems. I will discuss how traditionally kinematic techniques can be used to help define dynamic motion primitives that dramatically extend the capabilities of articulated robots operating in complex terrains. Specifically, I will present an approach for defining impedance-based primitives that enable different robotic platforms to blindly feel their way through unstructured two- and three-dimensional environments. Lastly, I will discuss future applications of my work in the areas of highly-adaptive soft and semi-soft robot locomotion and manipulation control.
Bio:
Matt Travers is a Project Scientist in the Robotics Institute at Carnegie Mellon, currently working in the Biorobotics Lab. He received his BS in Engineering Physics, as well as MS in Electrical and Computer Engineering from the University of Colorado at Boulder. Matt received his PhD from Northwestern University, working in the former LIMs Laboratory. His research interests include biologically-inspired robots, nonlinear estimation and control, dynamical systems, soft robots, and search and rescue robotics. Matt was nominated for a best paper award at the International Symposium on Safety, Security, and Rescue Robotics in 2015.
Matthew Travers : shape-based compliance in locomotion
Published on Jun 2, 2016
Matthew Travers
June 2, 2016
Abstract:
Natural systems move through unstructured, complex environments with an adaptability and robustness that motion-control scientists have yet to deeply understand, let alone reliably emulate. My work thus looks to biology, not to mimic or replay motions back on robotic devices, but to better understand the composition of successful motion-control strategies as they exist in the natural world. To this end, I will present recent work that uses modal decomposition techniques as well as tools from the field of geometric mechanics to better represent and ultimately interpret the motion of biological systems. I will discuss how traditionally kinematic techniques can be used to help define dynamic motion primitives that dramatically extend the capabilities of articulated robots operating in complex terrains. Specifically, I will present an approach for defining impedance-based primitives that enable different robotic platforms to blindly feel their way through unstructured two- and three-dimensional environments. Lastly, I will discuss future applications of my work in the areas of highly-adaptive soft and semi-soft robot locomotion and manipulation control.
Bio:
Matt Travers is a Project Scientist in the Robotics Institute at Carnegie Mellon, currently working in the Biorobotics Lab. He received his BS in Engineering Physics, as well as MS in Electrical and Computer Engineering from the University of Colorado at Boulder. Matt received his PhD from Northwestern University, working in the former LIMs Laboratory. His research interests include biologically-inspired robots, nonlinear estimation and control, dynamical systems, soft robots, and search and rescue robotics. Matt was nominated for a best paper award at the International Symposium on Safety, Security, and Rescue Robotics in 2015.